ITGAE Tool Ntwrk Ed

OPNET IT GURU: A TOOL FOR NETWORKING EDUCATION MSCIT Practicum Paper

REGIS UNIVERSITY

BY

Ranjan Kaparti

Professor Dan Likarish

2

TABLE OF CONTENTS Chapter 1: Introduction Page 3 Chapter 2: Solutions Page 6 Chapter 3: OPNET IT Guru Overview Page 9 Chapter 4: OPNET IT Guru for Education Page 17 Chapter 5: Multimedia Quickstart and Labs Page 21 References Page 41

3

Chapter 1: Introduction: The Problem

Computer networks are ubiquitous and such a state of affairs

creates demand for qualified personnel who can work with

networks. Unsurprisingly many institutions of learning (high-

school and higher learning) are teaching the relevant set of

networking technologies. Many are interested in creating new

programs and curricula with networking content.

The question arises as to how best to teach this material.

Michael Dixon, Tanya McGill, and Johann Karlsson point out that

many of the relevant concepts are difficult to really grasp in a

purely theoretical way. Lectures alone do not accomplish the

trick. And practical assignments - labs - must move beyond mere

programming assignments. Though writing say, a custom

communications protocol, is a good exercise, it falls short if

the goal is to teach the big picture of network communications

and associated high-level decision making and analysis. (Dixon,

McGill and Karlsson 71)

The nature of the technology itself is such that it is difficult

to get “hands on” with it from the point of view making the

technology available for classroom use. There is the issue of

expense. Many high schools for instance may not have the

resources to build an appropriate lab with all relevant hardware

and software. Furthermore real world networks may span not

4

merely a lab or two but rather entire buildings or even multiple

global locations. Noel Davis, Scot Ransbottom and Drew Hamilton

point out that:

Most networks are built to accommodate the needs of a single organization or group. Internetworking is a technology that accommodates multiple, diverse, underlying hardware by providing the means of interconnecting heterogeneous networks. Comparatively few individuals, academic institutions or corporations have networks exclusively dedicated for student use and experimentation. Prudent administrators limit student or general user access to operational networks. (Davis, Ransbottom and Hamilton 104)

Davis et al make some good points here. Even if there is a

network designated for student use it is rare that the network’s

sole purpose is to support the networking curriculum. Students

may gain valuable experience managing a network which is in

deployment (for general uses); however such experience has

limitations. For instance suppose a school decides to run the

Windows family of servers on their networks and the school

designates a lab for networking students’ use. Students may

learn a lot about Windows servers and associated technologies but

they will not obtain hands on training with Linux based

technologies. Even if one sets aside the worry about not being

able to learn the relevant range of technologies, students will

likely not be allowed to change settings and reconfigure things

to any significant extent in the Windows based network.

The bottom line seems to be that networking curricula face the

following dilemma: on the one hand it is expensive to set up a

networking lab and on the other, even if one makes that

5

investment, such labs have significant limitations from the point

of view of pedagogy. One cannot hope to cover the diversity of

technologies and configurations and furthermore large scale

networks cannot be built specifically for student use – the costs

are surely too prohibitive.

This problem of making available appropriate hands on computer

networking training is even more acute if one considers distance

education. In distance education (whether synchronous online,

asynchronous online, video based) students are in disparate

locations. It would be impossible to collect them together for

the sake of hands-on training – unless one mandates that for

brief periods of time – in which case it no longer would be

really purely distance learning. Indeed according to Brian

Cameron and K. Wijekumar:

Computer networking, as defined as the interconnection of computers and computing equipment using either wires or radio waves over small or large geographic areas (White, 1994), has long been regarded as one of the more difficult technology-related subjects to teach. Historically, this type of course was thought to require much hands-on interaction with the instructor and was not viewed as a good candidate for an online course. (Cameron and Wijekumar, 117)

It would be hard to imagine serious educational institutions

creating mini networking labs with actual hardware in every

distance learning student home or workplace. The costs, even if

partially to be passed on to the students, are simply too high.

6

Chapter 2: Solutions

Even though the picture prima facie appears bleak, there are some

viable solutions that would address not only the needs of

traditional but also those of non-traditional (distance)

networking students. The main solution of interest here is

network modeling and simulation software. In particular OPNET

IT Guru is going to be the focus.

Appropriate network modeling and simulation software could be

used to model networks down and up to whatever scale or detail

that is necessary. Students learn by creating virtual networks

and devices of all kinds and are not limited by network hardware

availability concerns. There is the prospect that distance

students can participate fully – all it would take is student

access to a workstation to run the software. The student’s home

computer itself would likely fit the bill.

Indeed studies have already been done that show significantly

improved learning outcomes in networking technology distance

students who used network modeling and simulation software and

those distance students who did not. Cameron and Wijekumar

claim that network simulation software increases student

motivation in online courses. One of the biggest problems in

distance education (according to them) is lack of sustained

student motivation and involvement. Network simulation software

7

gives students something tangible to do – a feeling of having

accomplishing something real and hands on and thus lessens drop

out rates. (Cameron and Wijekumar, 116f)

Cameron and Wijekumar actually conducted a study with an online

networking course for undergraduates (Cameron and Wijekumar,

118). One course used “static” network design package -

Microsoft Visio - for design assignments while the other course

used network simulation and modeling software for all their work.

It was not OPNET IT Guru but rather one made by NetCracker

Corporation. Nonetheless NetCracker’s software is dynamic as

well – permitting a range of modeling, design and simulation

functionality (LAN, MAN, WAN). It was found that students with

NetCracker software had a lower drop out rate, and rated their

learning to be much higher. They spent more time on their

assignments and explored “what if” scenarios. The ones with

Microsoft Visio felt that they did not quite know if their

network designs really worked or not before submitting them to

the instructor.

Cameron and Wijekumar further point to a number of studies which

collectively suggest the following good things about simulation

and modeling as a general pedagogical technique:

(a) “Simulations enable knowledge application through multidimensional problem solving.” (Cameron and Wijekumar, 119)

8

(b) “Simulations have been shown to improve knowledge transfer(Cameron and Wijekumar, 119)

(c) “Simulations have been shown to increase understanding of abstract concepts.” (Cameron and Wijekumar, 119)

(d) “Simulations and modeling can help instructors target students at multiple levels of learning.” (Cameron and Wijekumar, 119)

(e) “Simulations have the potential to improve students’ abilities with complex and evolving problem solving situations.” (Cameron and Wijekumar, 119)

It seems therefore to be almost a “no-brainer” to incorporate

network simulation and modeling software into the curriculum.

The advantages in terms of cost, student involvement and learning

outcomes, utility in distance learning, the ability to give

practical experiences about technologies of a wide range and/or

that are simply too expensive are so impressive and great that it

makes very little sense to reject this solution.

In the market there appear to be two major options for networking

and simulation software. Boson’s NetSim and OPNET’s IT Guru are

mentioned widely in the literature. The latter appears to be more

widely adopted in academy. The former seems to target the

Cisco’s line of networking examination and prep markets and not

necessarily networking in general (Boson Corporation, “Boson

NetSim: Education by Simulation”). In contrast IT Guru is

comprehensive and technology neutral in its capabilities and

versatility. However that fact by itself ought not to count

against NetSim. It may be that both pieces of software have

worthwhile roles to play in the curriculum. For the sake of

focus OPNET’s IT Guru will be pursued in depth here.

9

Chapter 3: OPNET IT Guru: Overview

IT Guru enables one to create a virtual network consisting of

relevant hardware, protocols, and application software (OPNET

Corporation, Slide 2 of “IT Guru QuickStart”, PowerPoint

Presentation, 2004). This network is a purely software entity

that can run on an individual workstation. Routers, switches,

web servers – almost anything found in real networks – can be

duplicated in an IT Guru virtual network. It can be scaled from

just a network of two workstations to one representing tens of

thousands running in a WAN.

Once a virtual network is created it can be manipulated in

various ways – for instance routers can be added or subtracted,

protocols switched around or altered, web servers added or

discarded – any permutation imaginable. The effects of various

alterations and diverse configurations can then be usefully and

quantifiably examined and analyzed. Importantly IT Guru allows

one to study and gather useful statistics about a virtual network

built from it. IT Guru permits not only the building of a

virtual network in software but also provides tools for

dynamically investigating the thus engendered network. (OPNET

Corporation, Slide 17 of “IT Guru QuickStart”, PowerPoint

Presentation, 2004).

10

The following high level architectural rendering of IT Guru is an

ideal place to get started. It shows IT Guru and its major

modules:

Source: OPNET Corporation, Slide 377 of “IT Guru QuickStart”, PowerPoint Presentation, 2004

The virtual network environment represents a network. It can

have a slew of components in every salient category. Opnet

defines a topology as a “collection of links, nodes and

configuration.” (OPNET Corporation, Slide 203 of “IT Guru

QuickStart”, PowerPoint Presentation, 2004) By “nodes” Opnet

means to include networking hardware of all kinds (routers,

workstations, switches, hubs etc). By “links” the underlying

connectivity technology (Ethernet, ATM, etc) and relevant

characteristics (latency, bandwidth) are meant. In

IT Guru Workflow – Inputs and Outputs

Virtual Network Environment

Traffic, Topology, and Configuration

Application Trace

Flow Analysis: Visualize and study steady-state data flow, capacity planning, failure analysis – extensive reporting

NetDoctor: Operational configuration validation – protocol-specific and policy-based DES: Simulates precise protocol effects for capacity planning, tuning protocol behavior, response

time engineering, technology migration, QoS, etc. ACE: Application performance troubleshooting: visualize performance, diagnose problems, explore

solutions

MVI, VNES, GUI

Sniffer, OPNET Application Capture Agent, etc.

Steady-state Throughput/Util; 100+ Reports; Path/Failure

Flow Analysis

Configuration validation NetDoctor

Graphs, web reports, charts, for all time-varying network performance metrics (e.g., app response time, queue depths,

ti t l

DES

Application Visualization, Diagnosis, Analysis, Protocol

ACE

11

“configuration” things like routing protocols, addressing,

whether VPN, are included. Some combination of nodes, links,

and configuration makes up a VNE. (OPNET Corporation, Slide 203

of “IT Guru QuickStart”, PowerPoint Presentation, 2004)

As the diagram shows there are several different ways of creating

an OPNET virtual network environment (VNE). The most

straightforward is by using OPNET’s GUI interface for that

purpose. One can select from a list of choices for all aspects

of a network. The general procedure is to select appropriate

hardware elements such as workstations, servers, routers,

switches, hubs, then connect them together according to the

desired topology. OPNET provides models of the major popular

brands of network communication hardware – Cisco for instance is

well represented with its family of routers and switches

available in the relevant OPNET IT Guru menus. (OPNET

Corporation, Slide 205, Slides 221-223 of “IT Guru QuickStart”,

PowerPoint Presentation, 2004)

The GUI method works well in many contexts surely but it can be

tedious if the network being modeled is elaborate and complex.

OPNET has made it possible to automate VNE creation via the VNE

Server technology. The VNE Server works in conjunction with

OPNET VNE clients to gather user specified network data. It

then has the ability to feed that data into IT Guru, producing a

virtual representation of the network. Clearly this is a highly

12

powerful method of engendering a VNE. It has the added

advantage of sidestepping user errors that are possible in a

manual VNE creation process.

Actually VNE server is powerful enough to process network

topology data from not only OPNET clients but also industry

standard network mapping and monitoring agents. For instance

VNE server can interface with HP Openview NNM, Cisco Netflow

Collector in addition to having the capacity to gather

information directly from various Cisco, Juniper and Nortel

switches and routers. It can also read from SNMP MIBs and in

the worst-case scenario perhaps ASCII files containing sundry

topology related information. (OPNET Corporation, Slide 207 of

“IT Guru QuickStart”, PowerPoint Presentation, 2004)

IT Guru can also utilize network device configuration file

information to create a VNE. Routers, for instance, have

associated configuration files that contain commands to the

router OS. Such files can be fed into IT Guru and IT Guru can

in turn generate topology information from them. The

restriction is that only Cisco and Juniper devices are currently

supported. (OPNET Corporation, Slide 208 of “IT Guru

QuickStart”, PowerPoint Presentation, 2004)

Opnet has devoted an entire specialized module called Application

Characterization Environment (ACE) for modeling and analyzing the

13

behavior of programs that reside above and are primarily users of

the network. In other words, these are above the topology layer

– they use network services but are not themselves strictly

speaking, part of the topology. Consider a web browser, or a

video conferencing client. Or a database query tool that talks

to a database over the network. Any and all such applications

can be modeled in IT Guru – one can make custom virtual

representations of such program and populate them inside a VNE.

Not only can IT Guru simulate network topology in the VNE but

also any high level application that runs on the network.

One must hasten to clarify that it is not that the entire

application’s logic and functionality will be duplicated inside

the VNE. Rather ACE (using appropriate Opnet agents) captures

an application’s communications and interactions with the

underlying network topology. This sort of snapshot of an

application’s network communications is called an application’s

“trace”. As an illustration of how ACE could be used, suppose

there is a new web application that remote users access on a DSL

connection. However they are experiencing more than expected

latency and delay. Suppose that ACE traces have been captured

both at the client and server. Now ACE, working in conjunction

with IT Guru, has the ability to simulate and analyze the entire

client server interaction here. The transactions between client

and server here can be simulated inside the appropriately

constructed VNE. ACE and IT Guru gives one the ability to

14

conduct detailed studies that would hopefully pin-point what the

trouble may be (or at the very least rule out problem spots).

For instance, it may be that the server is behaving correctly but

the client is firing requests at too rapid a rate or in a

redundant fashion. Such behavior, upon uncovering, may be fixed

by re-configuring or re-programming the client appropriately.

(OPNET Corporation, Slides 46-59, 176-178 of “IT Guru

QuickStart”, PowerPoint Presentation, 2004)

Flow Analysis, Net Doctor and Discrete Event Simulation (DES) are

IT Guru modules for all kinds of analysis and study of a

constructed VNE. In other words, these modules assist with

appropriately scrutinizing and playing with a virtual network.

They are not tools for generating a topology and associated

applications – rather they help in the further study of a given

VNE. They are the analysis arms of IT Guru. IT Guru enables

one to not only simulate a network but also to investigate such a

simulation in depth. The preceding modules are each designed to

probe different aspects of a VNE.

The Flow Analysis module enables conducting studies on network

traffic flows. In particular Opnet showcases report generation

on IP flows and ATM flows in permanent network connection (PVC)

scenarios. The module is capable of reporting on throughput,

and routing algorithm and traffic flow behaviors. Connection

failures are recognized and reported on. Flow Analysis would be

15

very useful in investigating what sort of impact changing network

configuration may have on general traffic flow in a network.

(OPNET Corporation, Slide 261 of “IT Guru QuickStart”, PowerPoint

Presentation, 2004)

The Net Doctor module can be used to check the configuration of

various nodes in a network model. For instance suppose that a

network’s routers are all supposed to behave a certain way.

After one captures this network into a VNE (possibly via the VNE

server) one can run Net Doctor to verify router configuration.

If Net Doctor says that the VNE routers are not working as they

ought to then that suggests that the real routers in the network

are not behaving as they ought to. Net Doctor can find issues

in routing, security, addressing and the like – one can program

Net Doctor with how things ought to look like and Net Doctor can

tell if things are as they should be in the VNE. Assuming that

the VNE is an accurate rendering of the relevant real network,

chances are that if there are problems with the VNE there are

problems in the real network. One can modify the VNE in various

ways to try and fix the issues – and then Net Doctor can be run

again. The final network configuration that passes Net Doctor’s

tests can then be implemented into the real network. (OPNET

Corporation, Slide 236-37 of “IT Guru QuickStart”, PowerPoint

Presentation, 2004)

16

Discrete Event Simulation (DES) module is in an important sense

near the heart of IT Guru. DES is the thing that fleshes out a

VNE in an operational sense. DES is the component responsible

for actually running a VNE in a live way. Once a VNE is set-up

it simply sits there inertly. DES is the component responsible

for actually running an appropriately configured VNE so that the

VNE begins to hum with life. DES is the simulator and analyzer.

As the traffic begins to flow around the VNE and all the virtual

nodes come online, DES runs statistical analyses of various kinds

on the VNE. The results of these analyses can be displayed in

various forms including in highly graphical ways - charts,

diagrams and so forth. Some examples are things like

throughputs, latency at various nodes, traffic volumes and

application response times – all sorts of relevant information

about the operations of the virtual network can be gathered up

for scrutiny. (OPNET Corporation, Slide 288-290 of “IT Guru

QuickStart”, PowerPoint Presentation, 2004)

17

Chapter 4: OPNET IT Guru for Education

It should be obvious from the preceding that IT Guru is an

industrial strength network simulation and analysis tool. And

indeed that is how OPNET is marketing it – it is not particularly

a product exclusively designed for use in the classroom. IT Guru

In other words IT Guru is not intrinsically for education nor has

it been created purely for classroom use. However OPNET

supports the use of IT Guru in the classroom primarily by making

IT Guru freely available for academic use.

IT Guru’s role in the classroom extends beyond merely learning

networking technologies. Learning IT Guru is itself a

worthwhile classroom objective because IT Guru is industrial

strength software with real world applications and career paths.

It is sort of like Microsoft Visual Studio – it can be used in

the classroom of course to teach programming with C# or C++.

But learning Visual Studio is more than simply learning

programming – in mastering it one is actually a step closer to

being a real-world programming professional. And Visual Studio

is hardly a tool designed purely for education.

Though IT Guru itself can be made the learning objective not very

many schools can afford to offer a course exclusively on it. IT

Guru’s educational potential will typically be that of teaching

networking technologies. The question then is how best to use

18

it to teach networking technologies. It has numerous modules

and several distinct methods of creating topologies. Which of

these modules and methods of VNE creation and subsequent

manipulation would work well for the purposes of pedagogy?

Prima facie the simplest approach for the run of the mill course

seems to be that of creating a VNE manually and manipulating it

via DES. Mapping a topology via automated methods (for example

with VNE server) might be too advanced for a regular course. If

the context is that of distant education it is entirely unlikely

that a distance education student would be able utilize VNE

server – for such an individual typically sits at a single

workstation possibly at home. In a non distance-learning

context, however, ACE and VNE server might make for exciting

components to an advanced level course. For instance imagine

advanced networking students generating the university’s topology

using ACE and VNE server. They could then manipulate the model

in various ways possibly in conjunction with the network

administrative staff. There are surely many possibilities.

At any rate since IT Guru is already adopted in some educational

institutions it is appropriate to briefly review such use. At

Murdoch University (Australia) IT Guru has been used in a network

communications course to teach client-server solutions paradigm.

In one lab students are asked to compare different client server

topologies from the point of view of network performance – they

19

start out with a 20 workstation network hooked to a single server

machine running FTP, E-Mail and Database servers. Then the FTP

server is placed on a different machine by itself. Finally the

FTP server is moved onto a separate wire. Via IT Guru’s

Discrete Event Simulation performance statistics analysis

students learn that the latter topology is the fastest. In

another lab, more complicated, students have to come up with the

best network configuration (hardware and software) for a small

business, given various constraints like costs. Here they design

a suitable VNE from scratch in IT Guru and run simulations with

DES. They then factor in costs for various configurations and

submit their final design along with all relevant work. (Dixon,

McGill and Karlsson 72-73)

At the Rochester Institute of Technology (New York) OPNET

simulation products are deployed in the regular classroom to

teach networking. There are several different ways in which it

is being used. In one type of lab students build token ring

and Ethernet networks with Opnet and compare performance under

various situations. In another, rather different, type of

context they use Opnet to design suitable networks for non-profit

organizations seeking to deploy networks for their organizational

benefit. In other words students are formed into groups to

consult for real world “clients” – these are typically non-

profits in need of network technology assistance – perhaps the

client is looking to institute an important upgrade or even build

20

a new network from scratch. Instructors act as resources to the

student groups and also review student work and recommendations.

Students have to work with the client to come up with

requirements, design, and finally implementation type documents.

Student use OPNET to justify and model optimum solutions.

(Perez-Hardy 94f)

Neither of the above institutions mentioned appear to be using

things like VNE server or even for that matter ACE. And they

have deployed OPNET in the regular classroom and not in a

distance format and so there is no issue of lack of lab resources

for mounting a VNE server. But just because these perhaps

advanced features of OPNET have not been utilized thus far does

not imply that they should not be so used by some curriculum in

the future. Certainly abroad – in Europe – some institutions

have incorporated application trace capture exercises in their

curricula. Recall that application traces can be imported via

ACE and this is indeed what the students do – they capture traces

of, say, their Usenet sessions, on their local computer. (J.

Theunis, B. Van den Broeck et. al. 3)

Regardless, in the remaining part of this work - very much in

line with current the majority general pedagogy - the focus will

be on manual topology building and subsequent DES based

simulation running using the topology. First a “getting

21

started” type tutorial and later sample IT Guru labs that could

be a starting point for a range of courses on networking.

22

Chapter 6: OPNET IT Guru Multimedia Quickstart and Labs

While it is possible to construct a tutorial with appropriate

screen captures and the like it may be more helpful to literally

demonstrate the basics of operating IT Guru using demonstration

multimedia tools. This tutorial can be viewed opening the web

page file called “SimpleSim.htm” – it is linked to a Flash file

called “SimpleSim.swf.” It shows how to create and run (using

DES) a very simple network topology – a single workstation hooked

to a server. Version 10.5 is the version of IT Guru shown in

the demo.

OPNET Labs

Once the decision is made to deploy IT Guru in support of a

course – distance learning or otherwise – the issue of precisely

how to leverage it arises. Given the sophistication of the

tool the range of choices for using it are vast. A range of

appropriate labs can be created on virtually any networking

topic. In the textbook world there is even a lab manual

available now – Computer Networks: Network Simulation Experiments

Manual – by Emad Aboelela. In this excellent work Aboelela has

hit a range of topics – from token ring to VPN – and the labs are

exclusively for IT Guru. However it may be that the manual is a

bit too sophisticated for the introductory to intermediate

23

networking student and beginning user of IT Guru. It is

certainly not for the neophyte.

In this paper a set of three labs is produced with appropriate

screen captures from OPNET IT Guru Version 10.5. They are

appropriate for someone who is at the introductory to

intermediate level with networking technology. They provide a

way of learning features of OPNET in addition to being hands on

review of some aspects of standard networking theory. They are

written in a generic way – they are intended to be worthwhile

anywhere from advanced high-school curriculum to graduate school.

It should be noted that IT Guru may be installed right on the

student’s own machine (OPNET IT Guru 9.5 is downloadable free for

educational use) or alternatively it may be mounted on a suitable

server or lab to which the student is given access – either

physically or online.

24

Lab 1: Hubs versus Switches in Small Business LAN Goal The goal of this lab is to compare performance of pure hub LAN with switched LAN in a tiny network. Overview We will create two scenarios. In one scenario there are five workstations that connect to a web server via a hub. Relevant performance statistics will be collected at both the workstations and the server. The same will be done in a second scenario except that instead of a hub it will be a switch facilitating the connection between server and workstations. IT Guru Step by Step Procedure Step 1. Launch IT Guru and create a new project – hub_switch_project. Step 2. Create a scenario – hub_scenario. Later we will add another scenario into the same project - switch_scenario. Step 3. Choose the default model library and default scenario dimensions. Step 4. Drag five Ethernet workstations and a Server into the construction area and connect them up with the aid of a hub. The topology should look like the screen capture below:

25

Step 5. Add profile and application nodes. Step 6. Open the attributes window for application node and select “Default” for the Application Definitions row. This makes all standard network applications available for use in our topology – HTTP, FTP, E-Mail, Database, etc. Step 7. We will create a custom profile – web application user profile – let us call the profile “web user.” To get there, go under the edit box – attributes window - for the profile. It looks as follows.

26

Go to profile configuration – it is highlighted above – and select “edit”. A new mini-window opens up – profile configuration table window – select row #1. It should look as follows:

Now under Profile Name change the name to “Web User”. Change the entry under Repeatability column to “unlimited”. Under Applications select edit and a new window pops out – this is the Applications Table window. Here we will select an application

27

that will be associated with the Web User profile. Select row 1 and pick “web browsing heavy.” It should look as follows:

Close all open windows. At this stage we have created a new profile called “Web User” and this profile can be applied in our topology. Step 8. Apply the profile “Web User” to all the workstations. This can be conveniently done by holding the control key down and selecting all five workstations and then making the appropriate change in the attribute window of any ONE workstation. One must make sure that the “Apply changes to selected objects” is checked on however in the attribute window of the workstation. The profile is applied by first opening the Applications tree. In the row entry for Application Supported Profiles select edit. A new window opens – Supported Profiles window – here go to row 1 and select Web User. It should look something like the following:

28

Step 9. Proceed to configuring the server. Select edit attributes, open the Applications tree and go to the Application Supported Services row. Select edit. A new window – the Applications Services Table window – opens up. Select row 1 and choose web browsing heavy. Essentially a generic HTTP server is set up on this machine at this stage. It may look as follows:

29

Step 10. Choose the statistics that ought to be collected during a simulation run. Under the DES menu bar select “choose individual statistics.” In the resulting window select the tree for “node statistics”. There are a number of options. Of particular interest for the purpose of this lab may be: Under Client HTTP: Object Response Time, Page Response Time, Traffic Sent and Traffic Received, User Cancelled Connections. Under Server HTTP: Load, Traffic Received, and Traffic Sent. Under TCP: Connection Aborts, Delay and Retransmission count. A screen capture of the DES statistics window looks like this:

Step 11. Run the simulation by choosing “Configure/Run DES simulation” under the DES menu. Step 12. Examine the results by selecting DES menu -> results -> view statistics. The statistics window might look as follows:

30

Step 13. To find a useful contrast let us now create another scenario within the same project. It is convenient to duplicate the entire topology for the hub scenario into the work area of the new scenario and subsequently make changes to the copy. Go under the scenario menu and select duplicate scenario. Name it “switch scenario.” A new work area opens up with a duplicate of the hub scenario. Select an Ethernet switch from the objects window (click on the open object palette button if it is not visible) and place it in lieu of the hub. Step 14. Run the switch scenario with the same statistics that were selected for the hub scenario. It is possible to compare the two statistics simultaneously – view it on the same graph – by selecting the All Scenarios option in the drop down list. Different colors are used for the different data. Examine all the performance data and answer the following questions. Lab Observations and Questions (1) In which scenario is there greater page loading delay? Explain why.

31

(2) Are there any worthwhile differences in any of the other statistics that were collected? Explain why one would expect differences or identical results. (3) Collect more statistics of your choice. Lab Concluding Remarks The lab should show that when there is a very small client/server network switches are worse than hubs from a performance point of view. This is explained due to latency involved in the switching process. The figure below is a screen capture showing page delay (points in red are from the switched scenario):

32

Lab 2: Hubs versus switches in medium scale LAN Goal The objective is to contrast the performance of hubs and switches in a medium scale LAN. Overview We will construct two scenarios – call them - hub_scenario and switch_scenario - under a new project called “medium_hub_switch”. The number of workstations will be at the medium scale – about 120 machines – each running web user profiles connecting to a web server. In one scenario the machines are hooked up exclusively to a hierarchy of hubs and bridges while in another a couple of switches come into play. IT Guru Step by Step Procedure Step 1. Create a new project called medium_hub_switch. Create an empty scenario called “hub_scenario.” Step 2. Choose a different model library than the default one by going to the object palate window and selecting the Ethernet model library from the drop down list. A new list of objects should appear. Step 3. We will create a subnet. Drag the subnet icon into the workspace. It should look as follows:

33

Step 4. Double click on the subnet icon – this will open up a new window – the subnet window for the subnet. Step 5. Create a new topology for the subnet using rapid network configuration. Go under the topology menu and select rapid configuration. Choose star. Choose Ethernet hub for the “center” and Ethernet workstations for the periphery. Make them 14 workstations. Select the 100 Base T connectivity link. It should look as follows:

Step 6. Click OK and IT Guru should automatically generate 14 workstations linked in star shape to a hub. Step 7. Create another 14 workstation linked to a hub topology in the subnet and place them alongside each other. (This can be done either by copying and pasting or by going through the rapid configuration menu again.) Step 8. Add an application profile icon to the workspace and create a profile called Web User (see Lab 1 for review of how to do this). Apply this profile to all the workstations in the subnet. Step 9. Drag in a bridge into the workspace and connect the two central hubs to the bridge. The result should look as follows:

34

Step 10. Return to the parent subnet window by clicking on the ‘Go to Parent Subnet’ button. Step 11. Copy the subnet icon that was just created and paste it at least three to four times – approximately 80-120 workstations in total now from all subnets. Step 12. Drag in a hub into the workspace and hook up the subnets to the hub via 100 Base T connections. During this process make sure to hook the hub to the bridge node in the subnet. In other words the link should be between the bridge in each subnet to the hub. Step 13. Link the hub to an Ethernet server. (Make sure the server acts as a HTTP server – drag in an application profile icon into the workspace to facilitate this – review Lab 1 for details.) Step 14. From DES menu select choose statistics and pick the page delay statistic for client HTTP. This is the key statistic for performance comparison between hubs versus switches. However in addition select any other statistic that may be of interest. Close the screen when done. The screen should look something like the following at this stage:

35

Step 15. Run the scenario. It may take several minutes under default settings. Step 16. There may be a number of warnings generated in the simulation log file to the effect that TCP connections were timed out. Changing the TCP parameter in the workstations to allow more than 3 reconnect attempts (the default) may dispel some warnings on a subsequent run of the simulation. However the underlying problem is network congestion! Step 17. Review the statistics collected. A number of nodes should have might have failed to secure connections with the server. There might be page delays running into the 10 seconds to 50 seconds range! Step 18. Duplicate the current scenario into a new scenario called switch_scenario. Step 19. Replace the bridges under each subnet with a switch. Step 20. Replace the hub in the top-level network with again a switch. The top level may look as follows – the center entity is the switch below:

36

Step 21. Select the same statistics as in hub_scenario – the client HTTP page delay statistic and run the simulation. It may take several minutes to run. Step 22. View the results. Compare the results. The red dots represent the switched scenario while the blue ones the hub/bridge scenario. A screenshot of the comparison is given below:

37

Lab Observation Questions and Mini-Labs for Further Study (1) In which scenario is there greater page loading delay? Explain why. (2) Are there any worthwhile differences in any of the other statistics that were (optionally) collected? Explain why one would expect differences or identical results. (3) Re-run the switch_scenario by replacing switches in the subnets with bridges in all the subnets. What happens to the performance then – to the page delay statistic? (4) Re-run the hub_scenario by replacing the central hub in the top-level network with a bridge. What happens to the performance then – to the page delay statistic? (5) Re-run the switch_scenario by replacing the top-level switch with a hub but retaining the switches in all subnets. What happens to the performance then – to the page delay statistic? (6) Write a report tabulating the different possible permutations. Extra Credit (7): Speculate on performance when the network scale increases to a 1000+ workstations. Build appropriate networks in IT Guru using techniques learned in this lab to test hypotheses. Lab Concluding Remarks Clearly as the number of workstations increases towards the 100 range the performance of hubs drops sharply – even when joined with bridges at the subnet level. But a bridge at the top-level (instead of the hub) improves performance dramatically. Switching on the other hand is effective in all configurations.

38

Lab 3: Viewing Simulation Routing Tables (Parts of this lab were inspired by the RIP Lab from: Aboelela, 77-88) Goal The objective in this lab is to simulate routing in a small WAN setting. Routing tables generated by the simulation will be exported to a file for viewing. The routing protocol in effect is RIP. Overview The idea is to set-up a campus LAN/WAN linked together by routers. The routers will run RIP. The routing table of a router will be exported to a file for examination. IT Guru Step by Step Procedures Step 1. Create a new project named “router_project”. Accept the default name for the scenario. Make sure to select “campus” for the scenario topology. Step 2. Drag in four “Ethernet4_slip8_gty” routers into the workspace (this is a model in the object palette called “internet toolbox” the default model palette. Step 3. Connect each router to the other in a ring using ppp_ds3 link. As in the following screen capture:

39

Step 4. Drag in the 100-T LAN icons into the workspace – there may be two for each router. Each router connects to two LANs – possibly in a separate building somewhere in campus. Each LAN icon represents a LAN – by default each contains 10 workstations – it may be configured at will by opening the relevant attribute window. It may look as follows (this topology is from Aboelela 80):

Step 5. Open the edit attributes window of any one router and expand Reports. In the RIP routing table export row select “Once at End of Simulation.” It may look as below:

40

Step 6. Run the simulation. Step 7. Open the results window and select the Discrete Events Table tab. Open up the tree node hierarchy until the routing table appears for the router. It may look as follows:

41

Lab Observation Questions and Mini-Labs for Further Study (1) What is the maximum hop count? Why? (2) What is the minimum hop count? Why? (3) Re-run the scenario after erasing one of the links from one router to another. What happens to the hop count? Lab Concluding Remarks In this lab we observed how a router running RIP generates routing tables. More advanced work along the same lines is possible by changing the router topology in various ways and by possibly failing links and routers mid-way during the simulation.

42

REFERENCES 1. Emad Aboelela, Computer Networks: Network Simulation Experiments Manual (San Francisco, CA: Morgan Kaufmann, 2003). 2. Boson Corporation, “Bosin NetSim: Education by Simulation,” Accessed May 4 2005 <http://http://www.boson.com/netsim/> 3. Brian H. Cameron and Kay Wijekumar, “The Effectiveness of Simulation in a Hybrid and On-line Networking Course,” SIGCSE’03, Proceedings of a Conference, February 19-23, Reno, Nevada, USA, 2003, Pages 117-119. 4. Xinjie Chang, “Network Simulations with OPNET” Proceedings of the 1999 Winter Simulation Conference, P. A. Farrington, H. B. Nembhard, D. T. Sturrock, and G. W. Evans, eds., 1999, Pages 307-314. 5. Capt. Noel Davis, Capt. Scot Ransbottom and Lt. Col. Drew Hamilton, “Teaching Computer Networks through Modeling,” Ada Letters, Sept~Oct 1998, Volume XVII, Number 5, Pages 104-110. 6. Michael W. Dixon, Tanya J. McGill and Johan M. Karlsson, “Using a Network Simulation Client-Server Package to Teach the Model,” Proceedings of a Conference, ITiCSE’97, Uppeala, Sweden, 1997, Pages 71-73. 7. Sylvia Perez-Hardy, “The Use of Network Simulation to Enhance Network Curriculum,” CITC4’03, Proceedings of a Conference, October 16–18, Lafayette, Indiana, USA, 2003, Pages 93-95. 8. OPNET Technologies, “IT Guru QuickStart” (PowerPoint Presentation), from OPNET Training Manual, (OPNET Technologies, Inc, 2004). 9. J. Theunis, B. Van den Broeck, P. Leys, J. Potemans1, E. Van Lil, A. Van de Capelle, “OPNET in Advanced Networking Education,” Accessed May 27 2005, <http://www.esat.kuleuven.ac.be/telemic/networking/opnetwork02_johan.pdf>